Home › Study guides › CCDV-F › Domain 5
CCDV-F · Domain 5 of 8 · 4 lessons · about 86 min
Domain 5: Model Selection and Optimization
How LLMs work in practice, the engineering behind SDK calls, choosing between Opus, Sonnet and Haiku, and managing tokens and cost. 16.8% of CCDV-F.
This domain is 16.8% of CCDV-F, about 9 of the 53 questions. It covers what a developer needs to know about the model itself and how to pay for it sensibly.
The vocabulary: a token is the unit a model reads and writes, and the context window is how many tokens it can consider at once. Sampling settings such as temperature make output more or less varied, which is why the same prompt can give different answers. Extended thinking and effort trade speed and cost for deeper reasoning. Prompt caching makes repeated input cheaper.
The recurring test: size the model and the settings to the task. The largest model with maximum thinking on every call wastes money and time; the smallest model on hard reasoning wastes quality. The right answer states the requirement (latency, cost, depth) and picks the lever that serves it.
What the exam guide tests
The official CCDV-F guide lists 4 skills for this domain. Exam questions are written against them, and so are the lessons: each row says where it is covered.
| # | Skill | Lesson |
|---|---|---|
| 5.1 | LLM Fundamentals (5.2%) | 5.1 |
| 5.2 | Technical Fundamentals (6.1%) | 5.2 |
| 5.3 | Model Selection and Tradeoffs (2.7%) | 5.3 |
| 5.4 | Cost and Token Management (2.8%) | 5.4 |
Lessons
- 5.1 LLM fundamentals: tokens, sampling, thinking and examplesHow tokens, context windows, sampling and next-token generation shape Claude's replies, and when to reach for thinking, effort, fast mode or examples.22 min
- 5.2 SDKs, REST and websockets: the plumbing around every Claude callWhat the Claude SDKs do on top of the REST API, when to call it directly, SSE versus websockets for the browser, and timeouts, retries and request ids.21 min
- 5.3 Choosing a model: capability, latency, cost and safe upgradesHow to match Haiku, Sonnet, Opus or Fable to each workload, which models think adaptively, and how to test a new model release before you switch.21 min
- 5.4 Cost and token management: model the bill, then cut itWhy a Claude bill outgrows its forecast, and how to budget tokens, track usage, model monthly cost and cut it with cache checkpoints and batches.22 min
Practice question
From the CCDV-F bank, tagged to this domain. Every answer option is explained. Nothing is stored, nothing to sign up for.
27 CCDV-F questions on this domain, free
The domain quiz in the question bank draws 10 random questions from the 27 tagged to Model Selection and Optimization, scores them and explains every option. Repeat it until the weak spots are gone, then sit the 53-question timed simulator.
Open the CCDV-F question bank → Start lesson 5.1 →
The question bank is free. It asks for an account only because the quiz engine has to store answers to score them and show which domains are weak. The questions on this page need nothing.